See.AI™ Metal Rebar Counting
Automating Rebar Counting for
Malaysia Steel Works
Malaysia Steel Works (KL) Bhd. — Masteel — was relying on manual headcount to track steel rebar bundles at QC, creating a significant production bottleneck. PixeVision deployed See.AI™ — a cloud-hosted web app accessed via smartphone — to automate counting entirely, boosting production output by 75% and cutting counting time by 80%. No hardware installation required.
LOCATION
Bukit Raja, Selangor
INDUSTRY
Steel / Metal Manufacturing
SOLUTION
Rebar Counting Software
CLIENT
Malaysia Steel Works (KL) Bhd. — Masteel
DELIVERY BOOST
COUNTING TIME REDUCTION
COMPLETE TRACEABILITY
THE CHALLENGES
Manual Counting Creating Costly QC Bottlenecks
Masteel’s production lines required workers to manually count steel rebar bars within each bundle at the QC stage before dispatch. This was done visually — workers physically counting individual bars, tallying on paper, and recording the results by hand.
The process was slow, error-prone, and created a significant bottleneck at the QC station. Miscounts led to customer complaints, inventory discrepancies, and shipment disputes. As production volume increased, the counting workload scaled with it — and so did the errors.
The factory had no digital traceability on bundle counts. If a customer disputed a quantity, there was no reliable audit trail to refer to. Every QC record was a paper sheet — which could be lost, damaged, or simply inaccurate.
THE SOLUTION
See.AI™ — Automated Rebar Counting
& Smart Analytics
PixeVision deployed See.AI™ as a cloud-hosted web application — no cameras installed, no hardware fitted. Workers simply open the web app on their own smartphones, point the phone camera at the rebar bundle from above, and See.AI™ instantly detects and counts each individual bar. The result is logged automatically to the cloud dashboard — accessible by management in real time.
Open Web App on Smartphone
The worker opens the See.AI™ web app on their own smartphone — no installation, no dedicated hardware. The app is cloud-hosted and accessible from any device with a browser.
Point Phone at Bundle
The worker holds their smartphone above the rebar bundle cross-section. The phone camera captures the image — See.AI™ processes it instantly via the cloud.
AI Bar Detection & Count
The AI model detects each individual rebar bar in the image and returns an accurate count within seconds — displayed on screen, no manual tally required.
Auto Cloud Record
Every count is automatically saved to the cloud dashboard with serial number, timestamp, operator ID, and bundle count — replacing paper records entirely and accessible by management in real time.
PROVEN OUTCOMES
Three Outcomes,
One Deployment
See.AI™ transformed the most labour-intensive part of Masteel’s QC process into a fully automated, cloud-connected operation — deployed entirely as a web app on existing smartphones, with zero infrastructure investment.
75% Production Boost
Processing more bundles per shift without adding headcount — the same team now handles significantly higher output because counting is no longer a bottleneck.
OPEX Savings & Labour Redeployment
Workers previously assigned to manual counting were redeployed to higher-value tasks. The cost of manual QC errors — customer disputes, recount labour — was eliminated.
ESG Compliance — 15,000 Sheets Eliminated
Paper-based QC records were replaced with digital logs. The reduction of 15,000 paper sheets annually supports Masteel's ESG reporting and commitment to sustainable operations.
Counting record on paper — slow, error-prone, no traceability, no digital record.
Smartphone web app — AI counting, cloud logging, full traceability. Zero installation.
Thousands of Paper Sheets Eliminated Annually
Full digital records replacing paper-based QC — supporting Masteel's ESG compliance reporting.
THE TRUTH - AI ACCURACY & ERROR HANDLING
What Happens:
When the AI Miscounts?
No AI system is 100% accurate — and we believe in being transparent about this. In high-volume production environments, edge cases occur: bundles with extreme bar density, partially obscured cross-sections, or non-standard stacking configurations can occasionally lead to a count that is slightly off.
The question is not whether miscounts can happen — it’s what the system does when they do.
See.AI™ does not claim to be infallible. What it does is detect its own uncertainty — automatically flagging low-confidence counts and prompting operators to verify, before the bundle leaves the QC station. The goal is not to eliminate human judgement, but to make it faster and more accurate when it matters.
AI Detects Low-Confidence Count
When See.AI™ identifies a bundle where detection confidence is below threshold — due to density, occlusion, or image quality — it automatically flags the result and prompts the operator to verify before approving.
AI Generates a Masked Recount Image
Rather than asking the operator to count the entire bundle from scratch, See.AI™ generates an annotated image — masking bars it has already counted so only the unverified bars remain visually prominent.
Operator Verifies & Approves
The operator reviews only the highlighted unmasked bars — confirming the remaining count in seconds. The final result is logged with a recount flag, maintaining a complete and honest audit trail for every bundle.
Recounting Made Fast — Not Frustrating
Traditional recounting means starting from zero. With See.AI™'s masking approach, operators verify only the uncertain bars — a handful, not the entire bundle. Recount completed in seconds, not minutes.
Ready to automate your production counting?
Tell us about your counting or QC challenge. See.AI™ is a cloud-hosted web app — no installation required, deployable on your team's existing smartphones.